# AiiDA Team Meeting 2023-07-13

###### tags: `team meetings`
###### time: 10am CET

[TOC]

### Present

* Sebastiaan
*

### Catch-up round

*Max. 3 minutes each*

Sebastiaan:
* [`BaseRestartWorkChain`: Factor out attachment of outputs ](https://github.com/aiidateam/aiida-core/pull/5983): Backwards compatible feature. Merging this soon.
* [Add `verdi profile setup`](https://github.com/aiidateam/aiida-core/pull/6023): Command that allows creating profile for any storage backend. Storage backend specific options are generated automatically (like with `verdi code create`). For the latter I used an ad-hoc custom spec to define the options. i know want to switch to `pydantic`. Anyone opposed to adding this as a direct dependency? For now against v1.10 because v2.0 is only just out and other depedencies downstream haven't updated yet. Big changes in the API.
* CJS: Should wait and move straight to v2: https://github.com/aiidateam/aiida-core/pull/6023#issuecomment-1626887517, someone can update sqlalchemy to v2 why you wait :wink: (the point being lets update the existing outdated dependencies, before introducing another one)
* Request for help on debugging: as soon as I added pydantic in the branch for the PR mentioned above, the tests started failing, but in a completely erratic manner. It is _only_ for Python 3.9 (v3.11 runs fine) and it has nothing to do with pydantic directly it seems. I can import it and use it just fine. It is just when I add 3 particular fields that things break. See the diff in [this commit](https://github.com/sphuber/aiida-core/commit/3830193ab689fae0c264d8596f32e40df6207d93). _With_ those three fields the test fail, and without them (commented out) the tests pass just fine. What's weirder, the code is not even really used yet, because I haven't added tests for it. And the tests are failing because the `stop_daemon_client` fixture runs into timeout errors, which eventually even starts breaking pytest itself. The daemon seems completely unrelated to the adding of a `pydantic.BaseModel` as class attribute to the `PsqlDosBackend`. I am at a loss and could use a fresh pair of eyes with some ideas :pray:
* Will also be on holiday next week and last week of July

Chris:
- Technically I'm booked on holiday until the end of the month, but still doing some work to tie things up and happy to meet
- continuing aiida-firecrest improvements; met with CSS on Tuesday to discuss upstream fixes/improvements
- Allowing for async file transfers (local <-> remote), can be reviewed:
- https://github.com/aiidateam/aiida-core/pull/6079
- https://github.com/aiidateam/plumpy/pull/272

Jason:
* [PR 6080](https://github.com/aiidateam/aiida-core/pull/6080) New docker image based on pure ubuntu and gracefully handle services ready to review.
* I push the image to ghcr.io, to test the PR image, just run: `docker run -it ghcr.io/aiidateam/aiida-core:pr-6080 bash`.

Giovanni:

- let's fix some time to discuss greenlets vs nest-asyncio?
- Will be on holidays until mid Aug
- Will have a PR to upgrade AiiDA to the new disk-objectstore on which I worked in the past 2 weeks (automatic compression heuristics, recompression while packing, pathlib support, 3.11 support, support for whence=2 and live unzipping into loose object when a packed object file is accessed in a random fashion, ...) - I need to make a release first, but yesterday I tested on 2 real containers (40GB and 80GB from me and Bonan, all good).
- Meeting to discuss design issues, e.g. asyncio for transport etc. - coding days

### Specific Agenda Items